K
Q

Secure communication between Ingress Controller (Traefik) and backend service on Kubernetes

4/13/2018

I'm trying to secure Nifi in a Kubernetes cluster, behind a Traefik proxy. Both are running as services in K8S. Traefik is secured with a public certificate. I want it to redirect calls to nifi, while securing the communication between Traefik (as an Ingress Controller) and the backend pods : Nifi.

Looks like the secure confiuration should lire in my Ingress YAML descriptor. Looks like I should issue a CA root to generate Nifi self signed certificate and load this CA Root in Traefik so it can validate the certificate sent by Nifi while handshaking with it.

But... I can't figure out 1) if this is the good approach, 2) how I can generate my stores (trust, ...) for NiFi using a CA Root, 3) how I should setup my YAML (insecureSkipVerify seems not to be supported, ...)

By advance, thanks for you help.

Cheers,

Olivier

-- Olivier
apache-nifi
kubernetes
kubernetes-ingress
proxy
traefik

Similar Questions

Kubernetes HPA gets wrong current value for a custom metric
Rancher 2.0 Agent installation
AWS VPC Peering connection issue
Unable to see join nodes in Kubernetes master

1 Answer

8/21/2018

I had the same problem and could solve it with the insecureSkipVerify flag.
The problem with traefik is, that NiFi gets the request from traefik and sends it's self signed certificate back to traefik for hand shaking. Traefik doesn't accept it, thus the handshake fails, leading to a bad_certificate exception in NiFi (has loglevel DEBUG, so you have to change the logback.xml file).

So one solution could be to add your self signed certificate to traefik, which is not possible at the moment, see this (currently) open issue.

Another solution, without 'insecuring' your existing traefik would be to add an nginx between traefik and NiFi. So traefik talk HTTP with nginx, which talks HTTPS with NiFi (this will be the next thing I'm trying).

Or you can set the insecureSkipVerify flag within traefik like I did in this daemonset.yaml:

apiVersion: extensions/v1beta1
kind: DaemonSet
metadata:
  creationTimestamp: 2018-06-21T16:18:46Z
  generation: 4
  labels:
    k8s-app: traefik-internal
    release: infrastructure
  name: traefik-internal
  namespace: infrastructure
  resourceVersion: "18860064"
  selfLink: /apis/extensions/v1beta1/namespaces/infrastructure/daemonsets/traefik-internal
  uid: c64a20e1-776e-11f8-be83-42010a9c0ff6
spec:
  revisionHistoryLimit: 10
  selector:
    matchLabels:
      k8s-app: traefik-internal
      name: traefik-internal
      release: infrastructure
  template:
    metadata:
      creationTimestamp: null
      labels:
        k8s-app: traefik-internal
        name: traefik-internal
        release: infrastructure
    spec:
      containers:
      - args:
        - --api
        - --ping
        - --defaultEntryPoints=http,https
        - --logLevel=INFO
        - --accessLog
        - --kubernetes
        - --kubernetes.ingressClass=traefik-internal
        - --metrics.prometheus=true
        - --entryPoints=Name:https Address::443 TLS:/certs/cert.pem,/certs/cert.key
          CA:/certs/clientca.pem
        - --entryPoints=Name:http Address::80 Redirect.EntryPoint:https
        - --insecureSkipVerify=true
        image: traefik:1.6.0-rc6-alpine
        imagePullPolicy: IfNotPresent
        name: traefik-internal
        resources: {}
        securityContext:
          privileged: true
        terminationMessagePath: /dev/termination-log
        terminationMessagePolicy: File
        volumeMounts:
        - mountPath: /certs
          name: traefik-internal-certs
          readOnly: true
      dnsPolicy: ClusterFirst
      restartPolicy: Always
      schedulerName: default-scheduler
      securityContext: {}
      serviceAccount: sa-traefik
      serviceAccountName: sa-traefik
      terminationGracePeriodSeconds: 60
      volumes:
      - name: traefik-internal-certs
        secret:
          defaultMode: 420
          secretName: traefik-internal
  templateGeneration: 4
  updateStrategy:
    rollingUpdate:
      maxUnavailable: 1
    type: RollingUpdate
status:
  currentNumberScheduled: 3
  desiredNumberScheduled: 3
  numberAvailable: 3
  numberMisscheduled: 0
  numberReady: 3
  observedGeneration: 4
  updatedNumberScheduled: 3

The insecureSkipVerify flag is changed within spec.containers.args.

Hope that helps!

-- Simon Linder
Source: StackOverflow